Welcome back to Good Better Best.
Figma just closed its first full quarter of charging for AI, and the numbers challenge the "seats are dead" consensus. This week I'm digging into why the seat may be one of the best containers for AI monetization — and what's worth stealing from Figma’s model.
Before we get there, next Thursday, we’re hosting Office Hours with Fynn Glover, CEO of Schematic to break down how to build trust around usage based pricing.
If you’re working on usage based pricing, and looking to improve predictability and transparency for customers, you won’t want to miss it. Register here →
🔌 PricingSaaS Partners power the next era of SaaS pricing
This Week in Pricing, Packaging, and Product
This week we observed 200+ changes. The highlights:
Mistral folded Education into Pro and bundled API credits into entry plans [Link]
Walnut added an AI credit allowance of 50K–200K by plan [Link]
UpKeep quantified Nova AI credit allocations across all plans [Link]
Lightfield ditched seat pricing for four credit tiers [Link]
Lindy scrapped its $8K assistant tier and cut Plus to $29.99/user [Link]
DeepSeek restructured API pricing into peak/off-peak billing [Link]
Fireworks raised on-demand GPU pricing 11–30% from Sept 1 [Link]
Together published pricing for Qwen3.8, Muse Glimmer, and DeepSeek V4 Pro [Link]
OpenRouter expanded its catalog to 500+ models across 80+ providers [Link]
Cursor dropped Composer from the "generous limits" benefit on Pro, Pro+, and Ultra [Link]
Suno added download caps and credit-to-song equivalents starting 9/3 [Link]
Runway upgraded Pro and Max to the Seedance 2.5 video model [Link]
HubSpot revealed pricing for additional Marketing Hub seats [Link]
Algolia renamed the Build plan to Free and cut records from 1M to 50K [Link]
Doodle raised Team pricing from $9 to $16 per seat [Link]
Grafana hid the Logs Query price ($0.003/GB) from the Pro plan card [Link]
Cloudflare removed the Network Services pricing table covering Magic Transit, WAN, and Firewall [Link]
JFrog cut developer capacity on the Ultimate Security Bundle [Link]
Fireflies added an Email Assistant AI feature to the Pro plan [Link]
Hiver launched an AI Agents add-on (contact sales) across all plans [Link]
Unbounce added MCP server support to the Starter plan [Link]
SafetyCulture rebranded to Mitti and restructured the Enterprise plan [Link]
Check out more updates on PricingSaaS →
Figma's Proof that Seats Aren't Dead
For two years the consensus has been that AI kills the seat. If the software does the work, why pay per human? Figma’s latest earnings call challenges that narrative.
Q2 is the first quarter where Figma has been enforcing AI credit limits, and the results suggest their seat model has been a powerful vehicle for AI monetization.
A few results worth highlighting:
More than 80% of paid customers above $10,000 in ARR were consuming AI credits weekly as of June 30.
Roughly two-thirds of that same cohort added Full seats at renewal.
Overall Net dollar retention was 136%.
Non-GAAP gross margin hit 85%, up 2.5 points sequentially — which CFO Praveer Melwani attributed directly to the first full quarter of AI credit monetization.
If you’re unfamiliar, here’s how Figma’s AI Monetization strategy works. Every paid seat comes with AI credits included.
Customers who blow through their allowance can buy add-on credit subscriptions or go pay-as-you-go. Three things are happening at once, and together they're why this is hard to beat.
1. It makes AI digestible. Bundling credits into the seat means the buyer's first experience with AI has no incremental cost decision attached. Nobody has to build a business case to try it. That's how you get 80% weekly adoption.
2. It makes AI predictable. Buyers hate the idea of an unbounded AI bill, and finance teams block deals over it. A seat with a known allowance is a number you can budget. The usage lever still exists — it just sits on top of a predictable floor.
3. It gives you a margin dial. Credits inside the seat let Figma decide exactly how much inference each price point buys. If costs move, the allowance moves. You are not stuck eating unlimited consumption on a flat fee, and you're not exposed to a customer who spikes 10x and torches your gross margin.
Escalating credit limits by seat type and PAYG fuel expansion as well. CEO Dylan Field described a familiar pattern where a core group of power users at a company will drive outsized usage, which opens up the rest of the org.
One large technology customer bought an AI credit add-on covering more than 25,000 paid seats, and engineers ended up holding more paid seats than designers. Another grew credit consumption 2.5x month over month after running power-user training.
Honest caveat: Figma hasn't disclosed AI credit revenue, attach rate, or what share of customers buy overages. And several new consumption surfaces — the agent, Code Layers, Motion — are still free in beta, with Figma eating the inference cost.
But the packaging lesson holds regardless. If you already have a seat model, it could be the perfect container for AI. Use tiered seats to make AI the reason to upgrade, put a meter on the top end, and let the credits do the expanding.
Thanks for reading! If you’re working on AI monetization and want to learn more about how we help, book time here.
Until next time,
Rob